Anthropic released Claude Opus 5.5 on September 22, 2026. In its announcement, Anthropic says Opus 5.5 “performs at the level of Claude Fable 5.1 on most work” and “generates output more than 30% faster than Opus 5.”
On this page
Availability and model ID
The model ID is claude-opus-5-5. Anthropic lists Opus 5.5 on the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure. On the product side it is available in Claude, Claude Code, Claude Code Enterprise, Claude Cowork, Claude Design, Claude Science, Claude Security, Claude in Chrome and Claude for Microsoft 365, among others.
Anthropic also offers a fast mode for Opus 5.5 with “up to 2.5x speed.”
Compared with Fable 5.1
Claude Fable 5.1, released on September 1, 2026, is the model Anthropic uses as the reference point for Opus 5.5. Besides the “level of Claude Fable 5.1” claim, the company says “the gap between Opus 5.5 and Claude Fable 5.1 is narrower than these scores suggest,” referring to its own benchmark table.
Preserved thinking and alignment
Opus 5.5 ships with a safeguard Anthropic first used on Fable 5.1. The company says: “Opus 5.5 is launching with preserved thinking, the anti-distillation safeguard we introduced with Fable 5.1. It stops API users from editing Claude’s prior context in an attempt to extract Claude’s reasoning.”
On alignment, Anthropic reports that on its automated behavioral audit “Opus 5.5 is the strongest-performing model we’ve tested to date,” and that it “scored better than any recent Claude model on nearly every measure of misaligned behavior.” These are results of Anthropic’s own internal audit.
Benchmark results
All scores below were published by Anthropic. Effort settings are those Anthropic lists for each result.
| Benchmark | Claude Opus 5.5 | Setting |
|---|---|---|
| Terminal-Bench 4.0 | 66.4% | Xhigh effort |
| FrontierCode v1.1 | 54.4% | Default (medium) effort |
| CursorBench 4.0 | 57.8% | Default (medium) effort |
| GDPval-AA v2.1 | 1846 Elo | Max effort |
| OSWorld 2.1 | 81.8% | Listed as “partial” |
| Chartography | 89.0% | With tools |
The results are reported at different effort levels, so the rows should not be read as one uniform configuration. Lab-published results for Opus 5.5 and other models are collected on our benchmark leaderboard.
What it changes
For developers on the API, Anthropic’s claim that Opus 5.5 is close to Fable 5.1 on most work makes it the obvious model to test against Fable 5.1 for the same tasks. The preserved thinking safeguard matters for integrations that manipulate conversation history: under Anthropic’s description, API users can no longer edit Claude’s prior context to extract its reasoning.
Opus 5.5 is listed on our AI model release timeline, and service health for the API is tracked on the Anthropic API status page.
